Introduction

Physical activity is important for good health, quality and longevity of life. The World Health Organisation (WHO) recommends that adults are physically active for at least 30 minutes daily and children at least 60 minutes daily (World Health Organisation, 2010). Epidemiological studies show that people who do not achieve these recommendations are at higher risk of developing hypertension, obesity, diabetes, cardiovascular diseases, certain forms for cancer and mental illnesses (Andersen et al., 2000). Life expectancy for inactive people has been estimated to be three to five years shorter than for physically active people; they use more healthcare services, have more frequent and longer absence periods from work and reduced health-related quality of life (Juel et al., 2008).

In modern societies it appears difficult to achieve a sufficient amount of physical activity for many people (Department of Health & Children Health Service Executive, 2014). Recent surveys have indicated that there are large proportions of inactive people particularly among children and young people, and this is a common trend in many countries (World Health Organisation, 2017). An active lifestyle accompanied with good dietary habits may be an effective way of avoiding overweight and obesity (Steinbeck, 2001). Unfortunately, the proportion of people who are overweight or obese has been increasing in many countries. This also applies for children and may pose a real threat for public health (Lobstein et al., 2004).

Public health interventions to prevent and reduce obesity are difficult to implement. Although short-term weight loss may be achieved through increased activity and restricted diet, long-term weight control is more challenging to achieve (Mead et al., 2017, Al-Khudairy et al., 2017). Similarly, structural interventions such as taxation policies, restrictive access to fatty  products, and easier access to infrastructures that encourage more active lifestyles (e.g. provision of safe bicycle lanes and walking paths) appear to be insufficient to change unhealthy behaviour (World Health Organisation, 2012). Individual-based interventions like “exerciseon- prescription” are costly and may provide only limited long-term effect (Sorensen et al., 2011). For this reason, it is obvious to consider how healthy life-styles can be encouraged at an early age as habits developed in early age have a greater chance of persisting into old age.

There are many opportunities to encourage healthy lifestyles in children and young adults (Mead et al., 2017, Al-Khudairy et al., 2017, Wang et al., 2015). Many clubs, charities and private organisations provide a wide range of opportunities for young people to engage in an active and healthy lifestyle. However, such organisations typically attract only certain
segments of the general population. Schools may have better and more effective means of reaching out to all children and support them with forming healthy habits. A wide range of
school-based programs have been developed worldwide, and many of them have been evaluated to document their ability to achieve recommended physical activity levels, weight control, and prevent or reduce obesity. 

However, school-based interventions have opportunity costs (Cawley, 2010), which implies that when a decision to provide school based intervention had been made, the resources cannot be used for something else. In other words, school time used on physical activity cannot be used for other types of teaching. This may not be a problem if the programme integrates physical activity into other teaching activities and achieve several teaching objectives. Evidence suggests that physically active children are more motivated and perceptive to traditional forms of teaching, so increased physical activity may actually improve the outcomes of the teaching (Haapala et al., 2017). The opportunity cost may be lower in school-based physical activities such as walking or biking to school, or activity during the breaks or afterschool. Nevertheless, it requires resources to integrate more physical activity into schools’ daily routines. These resources relate to schools, teachers, children and their families.

There exist comprehensive general frameworks for economic evaluation of physical activity and obesity prevention interventions (Edwards et al., 2013, Wolfenstetter, 2011). The general principles attempt to specify the costs related to the provision of the intervention and relate these costs to the major outcomes from the intervention. The outcomes relate to the children who become more active, and the improved health and other benefits that arise. To fully capture the costs and benefits of childhood obesity interventions, a long-term lifetime approach should be applied (Wolfenstetter and Wenig, 2011). Therefore, the use of modeling techniques in assessing this type of interventions may be unavoidable (Buxton et al., 1997).

The aim of this paper is to describe how economic evaluation can be applied to schoolbased interventions to prevent and reduce obesity. We use a recent Canadian study that promotes active living and healthy eating in schools as an illustrative case. 

Illustrative case study

This study assessed the long-term health and economic impacts of a school-based health promotion programme called APPLE Schools (Ekwaru et al., 2017). The intervention involved School Health Facilitators promoting healthy eating and active living among students, their parents, school staff and other stakeholders for a period of two years. These health facilitators were employed at ten “intervention” schools. The evaluation focused on ten-year old children and their two-year development in weight and height as compared to similar students from 148 randomly selected “control” schools (Fung et al., 2012). 

The economic evaluation of the intervention was devised as a long-term model-based assessment where the impact of the intervention was described in terms of changes in weight status, risk of chronic diseases, and quality-adjusted life years (QALY) for the cohort of 10-year olds until they reached the age of 85 years. Based on collected data for the intervention and control groups, two-year transition probabilities for three weight categories (normal, overweight and obese) were estimated. The two-year outcomes were based on observational data, and were expressed as change in weight categories due to the intervention.

To extrapolate these outcomes three scenarios were used to describe the weight development during the subsequent eight years after the end of the intervention. Children could either maintain the effect, they could continue with the improved effect for two more years, or the effect could decline by a specified annual rate. At the end of this eight-year period it was assumed that the weight category would remain unchanged from 18 years until 85 years. 

For the lifetime of the ten-year olds, health states with thirteen chronic diseases, no-chronic disease and the dead state were modelled for the three weight categories (total of 43 annual states). The empirical basis for these models included national population statistics (mortality), survey data (weight status), and epidemiological studies reporting on relative risk ratios, disease prevalence, and health related quality of life. This complex transition model was then used to estimate the incremental effect in terms of prevented life years with excess weight and chronic disease and gain in QALYs for the three scenarios relating to the weight development after the intervention. 

The study also considered the cost of the health facilitators for the intervention schools, and could therefore assess the cost-effectiveness of such interventions given a specified value of the included outcomes. This model-based approach enabled a wide range of sensitivity analyses of key assumptions applied in the model. 

Challenges for Economic Evaluation of Childhood Weight Interventions 

Although the study has generally high methodological quality, a number of methodological challenges for economic evaluations can be identified from the study. As with other types of models, economic models present a simplified description of reality imposed by a number of assumptions and input data obtained from various sources. 

The key assumptions applied when developing the model should reflect the cause–and-effect relations of the analysed interventions and the underlying biological process of the condition in question (Philips et al., 2006). The underlying conceptual assumptions for evaluating lifetime consequences of APPLE Schools were based on evidence from several longitudinal studies showing that childhood obesity tracts into adulthood (O’Brien et al., 2007). 

The effectiveness of the intervention was based on ecological change in the weight status distribution of the students attending APPLE Schools relative to those attending control schools at the follow-up time. The effect of the program at individual level was not determined. Ideally, data on effectiveness of an intervention should be based on a randomised controlled trial or a prospective cohort study with controls where the individual change in behavior and weight is observed for a longer period of time. With no empirical evidence on the long-term effects of a specific intervention, it is important to understand the motives and mechanisms for change and ability to maintain the new behaviour in order to make reasonable assumptions about the long-term effects. In the current study, three different sustainability scenarios were applied. The authors were not explicit about the assumptions made on the sustainability of the effect even though these assumptions are of crucial importance in regards to the study results. 

Ideally, the broad societal perspective including all costs and benefits no matter when and where they arise should be applied (Wolfenstetter and Wenig, 2011). However, a narrower focus is often taken to ease the complexity of the evaluation and to inform decision makers with specific budget responsibilities. Analyses with too narrow a perspective may, however, fail to  include important consequences, so both evaluators and users of evaluations should be aware of limitations of such methodological choices. In the present study, the school system’s cost perspective was applied, which was justified by the publicly funded education system in Canada. This choice led to an omission of some important aspects of the intervention, such as potentially reduced use of healthcare and social resources and increased contribution to the labour market as a result of intervention. Additionally, since the intervention involved engaging children, their parents, school staff and other stakeholders, it may have had social diffusion effects into other population groups, and these effects could also have been accounted for in the study. 

Another major challenge in the conduct of economic evaluations of preventive interventions is the appropriate identification and measurement of their benefits. In the described study, only the health gains in form of the reduced incidence of a specified set of obesity-related chronic diseases and mortality as a result of changes in weight were considered. However, positive changes in dietary and physical activity behaviors could be expected too. The health benefits that may arise from improvements in health awareness, self-efficacy and leadership as a result of prevented psychosocial consequences including low self-esteem, social alienation, discrimination and associated mental health disorders that can lead to a poor health-related quality of life (Davis et al., 1993, Davis and Christoffel, 1994, Davis et al., 1999) were not explicitly considered. 

Moreover, the underlying assumption that the intervention will produce health gains only if it results in changes in weight status can also be challenged. There is evidence showing that other pathways of improving health through increases in physical activity, independent of changes in weight, exist (Puder et al., 2011). On the other hand, there are also some health risks inherent in engaging in higher levels of physical activity in form of injuries (Warsh et al., 2010). Other ‘negative’ effects of the intervention may be the unrelated healthcare costs that may arise from extending people’s lives. These consequences of the intervention were not considered in the study.

Finally, applying a generic measure of health benefits including both the quality and the quantity of life lived in form of quality-adjusted life years (QALY) is a recommended strategy in economic evaluation of prevention interventions. However, there is no consensus on how healthrelated quality of life should be defined and measured in paediatric populations (De Civita et al., 2005). In the current study, for every year lived with excess weight, obesity or chronic disease, a decrement in health utility scores was assigned to the total QALY. The limitation of this approach is that the estimates of these decrements were obtained from two studies in which participants were 18 years of age or older, which might affect the validity of the estimated QALY.

Conclusion

Economic evaluations inform decision makers about the value for money of different interventions. Such information supports decisions to implement interventions that represent efficient use of resources. However, there are many methodological challenges for good evaluations. The case study described above illustrates some of these challenges. The case study results indicated that the Apple school program is successful in reducing obesity in children and that the long-term benefits in terms of improved health and less use of healthcare resources balance out the cost of providing the programme. This suggests that the Apple school programme in comparison with no programme is cost-effective and represent “good value for money”.